UCL Working Papers in Linguistics 10 (1998)
DP-Internal NP Movement*
TANMOY BHATTACHARYA
Abstract
In this paper I show that there are clause-like NP-movements within the DP. There have
been various proposals for N-movement within the DP (Bernstein 1993, Longobardi 1994
etc.) but none concerning NP-movement. I show that in Bangla (Bengali) at least two
types of DP-internal NP movements take place, both of which are due to specificity. The
analysis offered, in terms of a difference between nP and NP movement, additionally
derives the result that the deixis obtained in phrases where the NP has moved leftward is
non-locative.
1 NP ‘Object Shift’
In this section, I will briefly1 show that the first type of NP movement within the DP is
similar to the clause-like NP movement known as Object Shift in languages like Icelandic.
I will further show that this movement is based on phrasal specificity. Diesing (1992)
shows that indefinite subjects in Dutch and objects in Turkish allow a specific
interpretation of indefinite NPs in certain syntactic contexts. She equates specificity with
presupposition. For most authors, specificity essentially presumes an identified discourse
referent. Mahajan (1990) proposes to treat nominal specificity in Hindi/ Urdu as a
syntactic property. The core of the specificity effect (movement of the object NP out of
the VP) that Mahajan proposes can be imported into the type of theory that Diesing
discusses, which, in essence, forces all presuppositional material out of the VP at LF.
This proposal carries over directly to the following Bangla examples.
*
I am indebted to Rita Manzini, Ad Neeleman and Neil Smith for comments, criticism and suggestions on
earlier drafts of the paper; all remaining errors are mine. I am also thankful to two anonymous reviewers for
some crucial suggestions. Shorter versions of parts of the paper were presented at conferences at the
Universities of Manchester and York.
1
For details see Bhattacharya (forthcoming)
2
Bhattacharya
(1)
a. kal
Ek-Ta
lOmba chele eSechilo2
yesterday one-CLA
tall
boy came
‘a tall boy came yesterday’
b. kal chele-Ta
eSechilo
(presuppositional)
‘the boy came yesterday’
The nominal in the b example presupposes a prior discourse mention of its referent. It
shows that the shifted nominal chele ‘boy’ can receive a specific reading for the noun
phrase. The presuppositional nature of the nominal in (1b) is clear if we consider the fact
that (1b) alone can be an answer to a which NP question (2b) not (1a) (similar facts
observed by Kidwai (1995) for Hindi/ Urdu).
(2)
a. kal
ke
eSechilo?
yesterday who came
‘who came yesterday?’
b. kal
kon
chele-Ta
yeserday which boy-CLA
‘which boy came yesterday?’
eSechilo?
came
Presupposing the existence of some boy or boys, the question in (2b) targets the identity
of the boy involved. The data in (1) and (2) constitute the evidence for a presuppositional
analysis of DP-internal specificity in Bangla.
I now turn to the task of working out, within the syntax of the Bangla DP, the
mechanism responsible that gives rise to such specific interpretations.
Let us look at the distinction between the following.
(3)
2
a. oi
duTo
those
two-CLA
‘those two red books’
b. oi
lal
boi
those
red
book
‘those two red books’
lal
red
boi
book
duTo
two-CLA
(specific)
The transcription works as follows: T D R = Retroflex t, d, r; S = Palato-alveolar s; N = Velar n; E O =
mid vowels æ ⊃; M = Nasalisation.
DP-internal NP movement
3
The phrase in (3a) is the base order. Note, that the specificity effect is not clear from the
English gloss in (3). However, in the case of (3b) where the nominal boi has moved out of
its base position, the phrase is felicitous only if the nominal has a prior discourse reference.
I extend Diesing’s analysis here to NPs and suggest that a specific NP moves out of its
immediate nP-shell to a higher position. Notice that (3b) indicates a specific reading of the
NP lal boi ‘red books’. The N is specific or presuppositional in (3b) and therefore it must
move up. This leftward movement of the NP is shown in (4):
(4)
[DP ei [ClaP [NP lal boi] duTo tNP]]3
I propose that a feature of [specificity] on the Cla head (containing du-To as above)
ATTRACTS the NP to its spec as shown, giving rise to the specific reading. I call this NP
‘OBJECT’ SHIFT. I assume that when the [specificity] feature of Cla is not selected the DP
is non-specific as in (3a), similar to Collins and Thráinsson’s (1996) assumption regarding
the optional character of the strong N feature of Agro in Icelandic Object Shift at the
clausal level.
The NP movement above raises the question: What drives this movement? Since
movement in the framework adopted for this purpose (Chomsky 1995) is feature-driven,
the default option would be to formulate a mechanism for the movement observed in (4) in
terms of a feature. Let us assume that a filled Cla comes with an optional feature of
specificity in the numeration.4 Following Chomsky (1995: 277) we will assume that
optional features are added arbitrarily as the LI enters the numeration. In the present
theory, this option is exercised by the Cla in a specific DP in Bangla.
Again, essentially following Chomsky (1995: 281) since the features of the target which
enter into checking relations are non-interpretable, I assume that the Cla head bears a 3
I will modify the Bangla DP structure in (4) where the demonstrative (Dem) ei is shown to be a part of
the D-domain and instead suggest that the Dem is generated lower. In (3) Cla represents Classifier which is
the head of the middle ‘layer’ and may include either a Num (like du ‘two’ above) or a Q (like kOek ‘some’).
To keep the account as economic as possible, I will consider Cla as a complex head containing a Cla and
either a numeral or a quantifier.
4
An anonymous reviewer suggested that specificity be represented by a constant +Intp feature and the
marked word order for specific DPs be considered as a result of a strong +N feature on the same head. I
would like to suggest that since the specificity effect obtained in Bangla is configurational/ syntactic, it is best
represented as a -Intp feature which display, to quote Chomsky (1995: 278) “the special role of the property
of displacement of categories that is characteristic of human language”. Also, it avoids invoking another
feature like +N on the Cla head.
4
Bhattacharya
Interpretable syntactic feature of specificity and therefore, must be checked at the latest by
LF. To ensure that the checking occurs in overt syntax, I assume further that this optional
feature picked up by a Cla as it enters the numeration is STRONG. It is only when a Cla is
morphologically present, can a [specificity] feature be optionally selected. To see that this
is so, consider the following.
So far we have observed numeral expressions in Bangla carrying a Classifier Ta . There
are, however, certain classifierless Num-N sequences in the language. The following
examples are from Dasgupta (1983):
(5)
a. du deS-er
moitri
two country-GEN friendship
‘friendship between two countries’
b. tin
caka-r
gaRi
three
wheel-GEN
vehicle
‘three-wheeled vehicle’
c. tin
bOchor
three
year
‘three years’
d. car
paS
four
sides
‘Four sides’
e. tin
dik
three
direction
‘three directions’
A similar classifier-less Num-N sequence appears in the following measure expressions:
(6)
a. du gOj
two yard
b. tin
hat
three
hand/arm
‘three cubits’
c. car
miTar
four
metre
d. paMc
peala ca
five
cup
tea
DP-internal NP movement
5
Note that, crucially, in none of these phrases can the NP move leftward to give a specific
reading. The following, therefore, are not possible:
(7)
a.
b.
c.
d.
e.
f.
g.
h.
i.
*deS
duier moitri
*caka
tiner gaRi
*bOchor tin5
*paS
car
*diktin
*gOj
dui6
*hat#tin
*miTar car
*ca paMc peala7
A typical structure for the grammatical examples will be as follows:
(8)
[DP [ClaP car [NP paS ]]]
I take this data to mean that the Cla head in these cases, lacking a classifier, never
exercises the option of picking up a specificity feature when it enters the numeration. We
notice two things about these examples: that Cla lacks a classifier element, and that the NP
cannot prepose across the Cla. These two facts --- the absence of the classifier morpheme
and the absence of leftward NP movement -- must be correlated. The account that I
propose explains this correlation by giving the option of choosing a non-interpretable
formal feature of specificity for the computation only when the Cla is morphologically
present in the syntactic Cla-head.
Now consider the nature of this feature. Since I presume this feature to be noninterpretable, it must be checked in a Spec-Head configuration either in the covert or the
overt component. In the immediately preceding discussion, I have shown the need to
endow the Cla head of ClaP with a specificity feature as a lexical option made available if
5
The order is acceptable for a ‘vague’ meaning like three years or so; however, a vague meaning is far
from a specific meaning
6
7
The order in (7f) gives a vague meaning as in (7c)
That the order gives only a topicalised meaning (and not a specific meaning), is strengthened by the fact
that there is a distinct pause after tea
6
Bhattacharya
the Cla is morphologically present. Does this mean that the post-Cla NP must move
leftward whenever Cla occurs morphologically? No, for in the nonspecific DP examples in
(1a) and (3a), the post-Cla NP remains in-situ despite the presence of Cla. Choosing
among formal possibilities, I assume that in a given derivation, the option of assigning the
strong specificity feature to the Cla may or may not be exercised. But once such feature
assignment has taken place, there is no further choice. The complement NP must prepose
overtly to check this strong feature. This account assumes a standard mechanism to drive
the movement of NP to the Spec of ClaP. If, however, the numeration contains a
nonspecific Cla, then there is no need for feature checking and hence no overt preposing in
case of nonspecific (1a) and (3a).
Now consider the cases in (5) and (6). These DPs are without a classifier. According to
the analysis presented above, the Cla head in these DPs cannot carry any feature of
specificity. The impossibility of using classifiers with these expressions is a morphological
reflection of this fact. Why? Because the absence of classifiers precludes the choice of the
strong specificity feature for Cla. The point that (5, 6) help establish is that, in the absence
of classifiers, the Cla bears no attractor feature that could trigger complement NP
preposing.
This indicates that specific NPs cannot occur in Bangla, even though their features
would be interpretable, unless they enter into a checking relation with a specific classifier.
In section 3 where I investigate the nature of DP-internal deixis, I suggest that both
specificity and deixis are syntactic effects that obey a general Affect Criterion which has
been shown to include such diverse syntactic effects as Wh-criterion, Neg-criterion and
Focus-criterion. I argue that both (and not either) the spec and the head of a functional
projection need to be filled to obtain DP-internal specificity and deixis (see note 19).8
To conclude, what I have suggested is the following: If there is no morphological Cla,
there is no strong feature of specificity in the complex Cla head. If there is a morphological
Cla, then it has the option of picking up a strong non-interpretable feature of [specificity]
which then effects the leftward movement of the post-Cla NP.
8
This observation about the specificity of Bangla DPs which has a parallel in English WH-questions,
where a WH expression must check its feature even though interpretable, may be called the SPECIFICITY
CRITERION as suggested by an anonymous reviewer.
DP-internal NP movement
7
2 Kinship Inversion9
2.1 Introduction
In this section, I will present further evidence of leftward NP movement based on data -previously unnoticed in this or any related language -- involving kinship terms. I will show
that the marked word order NP-Possessive in (9) is due to KINSHIP INVERSION (KI), a
descriptive label for this movement. Unlike in other languages (e.g. in Italian, Longobardi
1994), in Bangla, I will show that the NP, rather than the N, moves up stranding the
possessive XP amar ‘mine’ in its base-generated position within the nP-shell10.
(9)
[DP [ClaP [NP
buRo baba] Cla] [nP amar [ tNP]]]
old
father
my-GEN
‘the old father of mine ...’
(KI)
I will claim that the landing site for this movements is also [Spec,ClaP]. That is, I will
argue that it is the specifier of ClaP (rather than the specifier of DP) which is the landing
site for specific NPs. However, before proceeding further, we need to sort out the structure
of the Bangla DP in relation to the position of the Poss and the Dem that will form the
basis of the discussion in the rest of the paper.
2.2 Possessives
I will begin the discussion by considering the position of the possessive (Poss) inside the
DP in Bangla. The structure of the DP in (4) and (8) provides an analysis for data like
(10a) where the Poss amar ‘mine’ is in [Spec,DP] and the Dem ei ‘this’ is in D as shown
in (11). (10b) shows that the Poss cannot be lower than the Dem in the structure:
9
A more detailed version of some of the ideas presented is discussed in Bhattacharya (1998a)
10
The presence of a nP-shell in the full-blown DP structure is justified in the next section (see (16)) and in
Bhattacharya (1998b).
8
Bhattacharya
ama-r11
ei
du-To boi
my-GEN
this two-CLA book
‘these two books of mine’
b. *ei amar du-To boi
(10)
a.
(11)
[DP amar [D ei] [ClaP du-To [NP boi]]]
The following phrases also support the structure in (11):
(12)
a. ram-er
gaRi
Ram-GEN
car
‘Ram’s car’
b. toma-r
du-To
lal boi
you-GEN
two-CLA
red book
‘your two red books’
c. rakhal-er
bORo bhai
Rakhal-GEN big brother
‘Rakhal’s elder brother’
So far we have seen that the Poss phrase occurs highest in the tree. But consider the
following:
(13)
a.
b.
c.
11
baba amar
father mine
chele amar
son mine
ma
mother
khub
very
khub
very
amar
mine
gorib!12
poor
duSTu!
naughty
SOt
manuS!
honest human
I will consider possessives as understood in well-known languages like English (e.g. John’s), and
possessive pronouns as similar and not different in Bangla in terms of their status and Case marking. Notice,
for example, that both John’s and my get the same Genitive Case marker -(e)r as in (i):
(i)a. jOn-er
b.
ama-r
‘John’s’
‘my’
12
A term of endearment may be added to the kinship term in these examples in a free gloss to convey the
sense of affection intended (indicated by the exclamation mark) when such phrases are used; e.g. dear
father, darling son etc
DP-internal NP movement
9
The data above shows that the normal, unmarked order of Poss-NP (as in 10 and 12) is
reversed with certain kinship expressions.13 Apart from the marked order of NP-Poss, the
data in (13) also shows that the Poss need not be highest in the tree. We have evidence
from other languages that the restriction to kinship terms is not unexpected. For example,
in Longobardi (1994), it is reported that in Italian, kinship terms have a cluster of
properties not shared by other common nouns. He suggests that kinship terms, in fact,
behave like proper nouns. Proper nouns in Longobardi’s theory obligatorily move to D.
Thus we can say that kinship Ns in Bangla are like proper names (as in Italian) and
therefore they move to D to give the order we notice above.
Bangla, however, differs from Italian in that in the former, it is the whole NP which
moves up. Consider the following:
(14)
a.
amar buRo baba khub bhalo
(unmarked Poss-NP order)
my old father very good
‘my old father is very good’
b. [NP buRo baba] amar tNP khub
bhalo (marked NP-Poss order)
(Lit.) ‘Old father (of) mine is very good!’
c. *baba amar buRo khub bhalo
The clause in (14b) above in contrast with (14c) clearly indicates that the N moves up
along with its modifiers14. In Bangla, therefore, it is a case of NP movement and not N (to
13
Notice that the use of this order is dictated by the stage or individual nature of the predicate in KI
cases if by Stage and Individual level we mean temporary and permanent properties of the subject
respectively:
(i) # baba amar aSben
(stage level)
father mine come-will
‘father mine will come’
(ii)
baba amar khub bhalo manuS
(individual level)
father mine very good man
‘father mine is a very good man’
This contrast is clearer in Hindi:
(iii)* baba mere aayenge
(stage level)
father mine come-will
(iv)
baba mere acche aadmii haiM
(individual level)
father mine good man is
14
I have provided some evidence in Bhattacharya (forthcoming) in support of [Spec,NP] status of Adj in
Bangla. However, even without such an assumption, the point about NP movement can be made by
10
Bhattacharya
D) movement as noticed in Italian and other languages. I will call this phenomenon
Kinship Inversion (KI). This will form the core of the discussion to follow. I will
specifically show that KI involves NP movement, that is, similar to DP-internal ‘Object’
Shift in 1 which also involved leftward NP movement.
Let us begin by considering the base and the derived position of the Poss in the DP
structure.
Since Poss is not highest in the tree, the only other XP position available in (4) or (11)
for the Poss is the [Spec,ClaP] position. Placing the Poss in [Spec,ClaP] creates more
problems than it solves. We have earlier proposed that a [specificity] feature of the Cla
head may attract a NP leftward to [Spec,ClaP]; therefore, the following, involving a Poss,
is also possible:
(15)
amar
boii
my
book
‘my two books’
du-To
two-CLA
ti
That is, the Poss amar ‘my’ precedes the moved specific noun boi ‘book’. Now, if we
claim that the Poss is generated (or even, ends up) at [Spec,ClaP] then the NP ‘Object’
Shift story is problematic unless we resort to multiple specs for ClaP. There is nothing in
principle to avoid generating multiple specs for ClaP but crucially, a Poss argument has
nothing to do with a Classifier Phrase, i.e. it does not ‘classify’ or quantify. There seems to
be no reason, therefore, to generate it at [Spec,ClaP]. Where does it come from, then?
According to some authors (Giorgi & Longobardi (1991), Mallén (1997) and others)
Poss is like an adjective (in some languages) and therefore must be generated within the
NP. In Lehman (1974), it is reported that attributive genitives in Old IE behave like
attributive adjectives but unlike the latter, are appositional in nature. This would suggest
generating the Poss in a pre-nominal position within the NP. If that is the case then in the
current framework, we can generate the Poss within the nP-shell as follows:
considering participial modifiers like the following:
(i) a. ghOre
bOSa baba
amar
home-LOC
sit-PPL father mine
(Lit.) ‘the home-sitting father mine!’
b.* baba amar ghOre
bOSa
father mine home-LOC
sit-PPL
(ib) is ungrammatical in the relevant phrasal sense; since the copula in the present is not expressed in
Bangla, it can have the clausal meaning ‘Father mine is sitting at home’. I thank John Payne for raising this
issue.
DP-internal NP movement
11
(16) [nP Poss [NP ADJ N]]
This would suggest that to derive a DP like (15), the Poss would need to move all the way
up to [Spec,DP] past the Cla head consisting of the Num-Cla complex duTo.15 In KI cases
as in (13), I will claim that the NP moves out of the nP stranding the Poss. KI, therefore, is
another instance of DP-internal NP movement. Two questions arise at this point: What
triggers this movement? and What is the landing site for the NP in KI?
Before we answer these questions, let us briefly discuss the position of the Dem in the
Bangla DP as it bears crucially upon the discussion to follow.
2.3 Demonstratives
2.3.1 Introduction. Consider the following data:
(17)
a.
ei du-To
lal
this two-CLA
red
‘these two red books’
b.* [lal boi]i
ei
du-To
c. ei [lal boi]i
du-To
‘these two red books’
boi
book
ti
ti
(specific)
The starred b phrase shows that leftward movement of the object NP across the Dem ei
‘this’ is barred. The crucial barrier here seems to be the Dem. If the Dem is a head, then it
is difficult to see how it can act as a barrier to XP movement. If the Dem is not a head then
it cannot occupy either D (our initial conclusion based on data as in 10, 12) or any other
head between D and Cla.
The proposal that the Dem may not be equated with D0 is well established in the
literature (Giusti (1997), Bernstein (1993), Brugé (1996) etc). They argue that the definite
article (at D0) and the Dem can co-occur in many languages:
15
Although nothing crucial in the present analysis depends on it, this movement is triggered by the
presence of a relevant feature ([Poss]) in D. In the case of KI (as in 13 and 14) this movement does not take
place in the overt syntax. I discuss this in detail in section 2.4.
12
(18)
Bhattacharya
a.
el libro este/ ese/ aquel
the book this/ that/ that
‘this book’
b. bäiat-ul
acesta (frumos)
boy-the
this nice
‘this nice boy’
(Spanish) Brugé (1996)
(Rumanian)Giusti (1997)
For (18b), Giusti suggests that N to D movement of bäiat ‘boy’ takes place across the Dem
as well as the Adj, if present. This shows that the Dem is neither at D, which has the article
-ul, nor in an intermediate head, otherwise the Head Movement Constraint (HMC) would
be violated for this local-N movement. Additionally, (18c) below shows that the Dem also
blocks AP movement to [Spec,DP] which is otherwise allowed in Rumanian.
(18)
c.
frumosul
nice
(*acesta)
this
bäiat
boy
The intermediate head position at whose spec the Dem is located, is needed as an escape
hatch for the N to D movement to proceed. We saw in (14) that in Bangla there is no N to
D movement. There is no compelling evidence, therefore, to posit a head X0 between D0
and Cla0; nevertheless the Dem behaves like an XP.
Some analyses on focusing at the clausal level posit a pre-verbal FP projection where the
head carries a feature of [focus] (Brody 1990). If we acknowledge the strong similarity
between clauses and phrases then we may want to posit a similar focus-like head in the
pre-ClaP position in the DP, that would provide us with a spec position for the Dem. Let
us see if there is any independent evidence for doing so, i.e. is there any evidence for a
Focus head in the DP? Giusti (1996), based on data from Albanian, proposes a FocP
within the DP.16 However, I will differ from Giusti in terms of the position of the Dem
16
Giusti’s suggestion is based on the following. In Albanian, Adjs appear postnominally (i) but can be
prenominal if the Adj is emphasised as in (ii):
(i) një grua e
bukur
a
woman the
nice
‘a nice woman’
(ii) një e
bukur
grua
‘a NICE woman’
Giusti considers this to be due to a movement of the AP to [spec,FP]. I have extended this proposal for a
FocP inside the DP to include contrastive focus as well as the following discussion will show (see section
3.1).
DP-internal NP movement
13
(which is at [Spec,DP] for her) and suggest that the Dem is merged at the specifier of the
Focus Phrase. My reasons for proposing a FP inside the DP in Bangla are different from
the ones proposed for Albanian (or Russian) in Giusti. I will provide empirical evidence
from Old IE and Bangla and theoretically, I will suggest that a DP-internal Focus position
derives the nature of deixis more economically. I will specifically propose that deixis in the
Bangla DP is obtained through the interaction of the Focus head with the lower head Cla
(see section 3).
2.3.2 DP-Internal Focus. Given the theoretical assumptions proposing a functional
projection for the Dem and given what we have seen so far for Bangla, let us propose an
FP within the Bangla DP. The F head is silent in Bangla (as in other languages mentioned
in Bernstein (1997) who also proposes a functional projection, not Focus though, below
the level of DP) but carries a strong [focus] feature, if Selected, that must be erased before
the derivation reaches spell-out. The new structure for the Bangla DP is given in (19). That
is, we have generated the Dem as a specifier of the intermediate FP projection. There are
several historical studies (Peterson 1930, Hamp 1980, Hazelkorn 1983 and Sheilds 1994)
of IE personal pronouns supporting a focus analysis of Dems in general. These studies
broadly conclude that the deictic particles attached to IE pronouns were “emphasising”
particles. I argue that Dems in Bangla are derived from personal pronouns plus the particle
-i.17
17
Notice that the augment -i used to form Dems like ei, oi, Sei ‘this (proximal), that (distal), that
(sequent)’ from personal pronouns like e, o, Se is homophonous with the emphatic particle –i in
Bangla:
(i)
a.
rajen-i
baRi jabe
Rajen-EMP
home go-will
‘Only Rajen/ Rajen himself will go home’
b.
rajen baRi-i jabe
‘Rajen will go to the house itself’
c.
rajen baRi jabe-i
‘Rajen will definitely go home’
14
Bhattacharya
(19)
DP
Spec
D’
D
FP
Spec
ei
F’
F
ClaP
Spec
Cla’
Cla
du-To
NP
ADJ
lal
N
boi
The new DP structure now explains the inability of the NP to move across Dem in (17b)
in terms of minimality. Let us now see how the above structure can derive the simple DP
in (20):
(20)
ei
du-To
boi
this
two-CLA
book
‘these two books (here)’
I suggest that a functional head F with a strong [focus] feature is selected from the
Numeration after merging of the Cla duTo and the NP boi has taken place. This strong
feature induces raising of the Cla head to F and thus ‘Dem-hood’ is established. I am
assuming that the Dem ei is merged at [Spec,FP]. This is shown in (21):
(21)
[FP ei [F duTo] [ClaP [Cla duTo] [NP boi]]]
Deixis is obtained by achieving a certain syntactic configuration and is not a lexical or a
featural property of a particular lexical item. We will utilise this concept of deixis in our
analysis in section 3.2.
Let us take stock of the developments so far. We started with a quest to chart out the
movement of the Poss in examples like (10). We rejected the possibility of generating it at
[Spec,ClaP] and went on to reassess the three-tiered DP structure proposed in (4, 11). We
DP-internal NP movement
15
concluded, by looking at the behaviour of the Dem in Bangla, that a three-tiered structure
can no longer be maintained and that we need to posit a head (of which the Dem is the
specifier) between the D and Cla as in (19). As it seems reasonable that the Poss is
generated at a lower shell and moved to a Case position, let us look at the DP that we
started with. However, looking at (10) again, repeated here as (22), the structure above
seems inadequate as the Dem as an XP would block the movement of the Poss XP across
it. This is clearly not the case in (22b):
(22)
a.
ama-r
ei
du-To boi
my-GEN
this two-CLA book
‘these two books of mine’
b. *ei amar du-To boi
We have already seen that the Dem cannot be a head and that it is generated as the spec of
FP. There is no reason to assume that the Poss is a head since it is a full NP with genitive
Case. We have also seen that the Dem blocks movement of other NPs across it (17) but
not the Poss (as above). Is there a way out of this?
Given either the MLC (Chomsky 1995) or the scopal MLC of Manzini (1998), shown
below (23), movement of the Poss to [Spec,DP] is not a problem since the Dem is not an
attractor of the [poss] feature.
(23)
Given an attractor feature F and an attractee feature AF, F attracts AF only down
to the next attractor F’ for AF
The Poss in our theory moves up to [Spec,DP] to check the [poss] featue – presumably a
Case feature -- at D. This checking takes place overtly in case of overt Poss movement to
[Spec,DP] as in (22a), or after spell-out in case of KI when the Poss is stranded in its basegenerated position. This may be considered similar to Kayne’s (1994) suggestion that in
English the Poss NP (as in John’s) raises past an empty D0 head at LF. Overt/ covert
movement for Case checking is reflected on the obligatory presence of Genitive Case on
Poss in all cases.
2.4 KI trigger
Having decided the base position of the Poss and Dem in the last two sections, we are now
16
Bhattacharya
in a position to answer the questions raised immediately below (16). KI involves shifting
the kinship NP to the left of the Poss in case of certain predicates (see note 13). I will now
proceed to show that a feature of the Cla head induces the inversion noticed with kinship
terms. First, the following pair shows that inversion is obligatory when an ‘affectionate’
Ti instead of the regular –Ta is used with kinship terms:
(24)
a.
bon-Tiamar khub Sada-Sidhe
sister-CLA my very plain-straight
‘sister mine is very plain and simple’
b. * amar bon-Ti khub Sada-Sidhe
This shows that –Ti induces KI, i.e. both the use of this particular Cla and KI have
matching requirements. More importantly, the Cla morpheme instantiates a feature to the
Cla head which is responsible for this inversion. Secondly, the following contrast shows
that kinship terms when associated with Proper Names do not undergo KI, but may only
do so in the presence of a Cla.
(25)
a.
rakhal-er
bhai/ *bhai
rakhal-er
Rakhal-GEN brother/ brother Rakhal-GEN
‘Rakhal’s bother is very good’
b. bhai-Ti rakhal-er khub bhalo
‘the brother of Rakhal is very good’
khub bhalo
very good
This example again show that a feature of the Cla head (instantiated by the Cla morpheme)
is responsible for KI in (25b).
Now that we know that a certain feature of the Cla head is responsible for KI, it is also
reasonable to assume that the feature of [specificity], responsible for NP- ‘Object’ Shift,
may be involved in case of KI. The fact that proper names do not undergo KI without the
overt presence of the Cla may have something to do with the base position of such nouns
given that their semantic status is different from common nouns, however, we do not deal
with issue further in this paper.
In (26), the reading obtained is specific, i.e. something is being said about two specific
brothers.
(26)
bhai
duTo
amar, ...
brother two-CLA
mine
‘brothers two (of) mine, ....’
DP-internal NP movement
17
KI in the case of (26) therefore must involve raising of the NP bhai through [Spec,ClaP]
inducing the observed specificity. On the surface, there is no way to distinguish NP
movement to [Spec,ClaP] or further up to [Spec,DP] if there is no other head present in
between. Investigating this further will lead us to answering the second question raised in
the paragraph following (16), namely, What is the landing site for the NP in KI?
2.5 Landing site for the Kinship NP
In connection with (16), I mentioned that the Poss in KI is stranded in its base position
within the nP shell. The Poss in such cases, therefore, does not check the [poss] feature at
[Spec,DP]. Let us consider the following data in this connection:
(27)
amar
BON khub bhalo, bhai-Ta-i
bOjjat
my
sister very good brother-CLA-EMP
nasty
‘my SISTER is very good, it’s only the brother who’s nasty!’
The focus on bon ‘sister’ now picks out bon as opposed to other objects that may belong to
the set already created by amar.18 Let us now consider the following where the Poss is
focused:
(28)
AMAR bon
khub bhalo, (tomar-Ta bojjat)
my
sister very good your-CLA nasty
‘MY sister is very good, (it’s yours who is nasty!)’
Comparing (27) and (28), we find that in the unmarked Poss-NP order, both the possessed
and the possessor can be focused. Let us now see if this is the case for the other order we
have been looking at, that is, the marked order of NP-Poss in KI. The inversion, I suggest,
breaks up the relation normally obtaining between the Poss and the NP. That is, if there is
a θ-role of POSSESSION which normally obtains between the possessor and the possessed,
it is unavailable in case of KI. I will assume that in a typical KI case as in (13), the
thematic properties of the Poss are satisfied in the covert component. In the overt syntax,
18
In the theory of Rooth (1985), focusing creates several alternate sets one of which is then picked up by
the denotation of the NP.
18
Bhattacharya
therefore, the Poss remains in its merged position during the derivation.
Let us see the effect of focusing on the marked order:
(29)
a.
CHELEi
amar ti
khub bhalo, ...
son
mine
very good, ...
‘SON mine is very good, ...’
b.* chelei AMAR
ti
khub bhalo, ...
We conclude that in the marked order of NP-Poss, the Poss cannot be focused. This
empirical conclusion is imported into our analysis of KI (see section 3.2 for details) where
NP movement leaves the Poss stranded. Does this mean that the NP, then, raises all the
way up to [Spec,DP] in these cases? If it did, then the following data will be problematic.
(30)
a.
ei bhai
amar
this brother
mine
b.* bhai ei
amar
If the kinship noun bhai ‘brother’ raises to [Spec,DP] to check some feature on the D, the
unacceptability of (30b) is strange. It is, therefore likely that the NP does not raise all the
way up but only up to [Spec,ClaP]. Therefore, in (30a) the NP bhai moves to [Spec,ClaP].
This is one reason to analyse KI as NP movement. However, any movement to this
position has been analysed so far as movement due to specificity. Both (26) and (30)
therefore indicate that the landing site for the NP in KI is [Spec,ClaP] and is brought about
by the feature of [specificity] responsible for NP- ‘object’ Shift discussed in section 1.
3 Deixis inside the DP
3.1 Introduction
In this section, I will investigate the nature of the deixis which obtains inside the DP and
show that DP-internal NP movement leaves similar deictic results for both ‘Object’ Shift
and KI. I will claim that the analysis offered here obtains the DP-internal deictic facts for
free. However, as a result of this investigation, I hope to show that although on surface
these two types of NP-movements look similar, they may involve moving different
“chunks” of the NP-shell.
Bernstein (1997) points out the difference between the pair in (31) by suggesting that in
DP-internal NP movement
19
(31a), the deictic effect is obtained by moving the Dem to D0, this movement does not take
place in the syntax for (31b).
(31)
a.
b.
this woman (right here)
= this woman
this woman (from Paris)
= a woman
(Bernstein’s (19))
(deictic)
(indefinite specific)
That is, for Bernstein, deixis is obtained through a movement of the Dem to D0 whereas in
the case of the indefinite there is no movement of the Dem.
Bernstein’s account of deixis is unsatisfactory on at least one ground. She proposes
movement of an XP (Dem) to an X (D0), but she does not specify what feature triggers this
process. Based on empirical evidence, I will suggest that deixis is not obtained through
movement to D0 but rather of a lower head into the head of which the Dem is a specifier,
FP in our formulation. I will suggest that the deictic effect in (32a) is obtained through the
Dem being merged at [Spec,FP] and some lower head moving to F (also see 21).19
(32)
a.
ei du-To
boi
this two-CLA
book
‘these two books (here)’
b. ei boii
du-To ti
‘these two books’
(deictic)
(specific)
The data shows that in (32a) the deictic meaning is more important (shown in the
translation by here), whereas in (32b) the specificity of the books is more important. In
(32a), I will assume that the Cla has head moved into F resulting in the kind of deixis
obtained. An absence of this head-movement in (32b) results in, what I will call, ‘NONLOCATIVE’ deixis. Let us look at the nature of DP-internal deixis in greater detail.
19
Effectively such a condition is similar to the Focus Criterion of Brody (1990):
(i) a. A [+F] operator must be in spec-head agreement with a [+F] X0
b. A [+F] X0 must be in a speck-head agreement with a [+F] operator
20
Bhattacharya
3.2 NP Movement and deixis
In this section I will investigate the effect of DP-internal NP movement on the nature of
deixis inside the DP. Particularly, I will show that NP movement makes the deixis of the
phrase non-locative in a sense to be made precise. The difference in deixis will also lead us
to propose a distinction between two types of NP movements (section 3.2).
The pattern obtained in (32) is reflected in a larger set of data involving the Dem and the
Poss. In connection with (33), I will claim that whenever there is a Poss preceding the
Dem, the nature of deixis obtained is contrastive (as in 33a). That is, apart from
performing its deictic function of ‘pointing’, the Dem seems to contribute a contrastive
meaning to the phrase as well. The deixis obtained in (33b) is non-locative. That is, there is
no ‘here and now’ interpretation of the Dem in the latter.
(33)
a.
ama-r
ei
du-To chele
my-GEN
this two-CLA son
‘these two sons of mine’
b. ama-r
ei
chele du-To
my-GEN
this son
two-CLA
20
‘these two sons of mine’
(contrastive)
(non-locative)
The contrastive meaning of (33a) is suggested by the full sentence (34) where the Dem
clearly contrasts the set of two boys with another set consisting of Rakhal.
(34) ama-r ei du-To
chele khub bhalo, rakhal-Ta-i
boka
my
this two-CLA
son
very good Rakhal-CLA-EMP
foolish
‘these two sons of mine are very good, it’s only Rakhal who’s a fool’
That is, the set of two sons is contrasted against another consisting of Rakhal. In (34) ei
duTo chele is the constituent in focus which is set in contrast with the phrase appearing
within the negative contrastive adjunct -- a necessary and sufficient diagnostic for focus in
Rochemont (1998). Notice that the NP has not moved across the Cla duTo, that is, there is
no ‘object’ shift involved here. Note further that the Dem in (32a) is used deictically
whereas in its specific counterpart (32b), the Dem loses its deictic effect. We note that
20
The similarity of the English gloss indicates the impossibility of the boys two order for specific two boys.
However, the fact that the natural focus for this expression in English is carried by these confirms the
contrastive nature of the deixis obtained for the Bangla equivalent.
DP-internal NP movement
21
specificity overrides deixis inside the DP. Crucially, leftward NP movement involved in
NP ‘Object’ Shift renders the deictic effect of the Dem ‘non-locative’.
Similarly, in the case of NP movement due to KI, the resulting deixis of the phrase is
non-locative. This is shown in (35).
(35)
a.
ei du-To
bhai
amar
this two-CLA
brother mine
‘these two brothers (of) mine, ....’
b. ei bhai
du-To amar
‘these two brothers (of) mine, ....’
....
(contrastive)
(non-locative)
In the next section we will show that the difference between the two DPs in (35) is one of
specificity. This would entail movement of the NP to [Spec,ClaP] in the case of the
specific version (35b). I will also show that the type of DP-internal deixis obtained with
NP movement is predictable from the analysis.
3.3 nP/ NP Movement
3.3.1 Introduction. In this section, I will first show that the grammar of the Bangla DP
must distinguish between at least two types NP movements. In other words, I will provide
well-motivated reasons to move different “chunks” of the nP within the DP. This
distinction will be shown to necessitate generating the F head. In the course of the
discussion we will also see that the type of deixis obtained falls out as a consequence of
the analysis.
To get a quick overview of what we are going to propose, let us look at the following:
(36)
a.
amar ei
bhai
duTo
my this brother
two-CLA
‘these two brothers of mine’
b. ei bhai
duTo amar …
‘these two brothers of mine …’
(NP‘O’S; nP movement)
(KI; NP movement)
That is, these two operations, NP ‘Object’ Shift and KI, in effect, result in different
‘chunks’ of NPs being moved. As expected, the deicticity effect obtained in both (36a) and
(36b) is non-locative, since both NP‘O’S and KI induce non-locative deixis (section 3.1).
22
Bhattacharya
However, in their non-specific counterparts, the order of things is different:
(37)
a.
amar ei
duTo
bhai (nonspecific; Poss movement)
my this two-CLA
brother
‘these two brothers of mine’
b. ei duTo bhai amar …
(nonspecific KI; NP movement)
‘these two brothers of mine …’
I will suggest that the deixis obtained in both a and b of (37) is contrastive. Notice that in
both cases nothing intervenes between the empty F head and duTo, the Cla head. The Cla,
therefore, can head move to F in each case, in effect, deriving the contrastive deixis
obtained. I claim this to be the theoretical justification for a head between D and Cla.
3.3.2 nP Movement. In this section I will propose that movement due to specificity is nP
movement to [Spec,ClaP] followed by a movement of the Poss to [Spec,DP] if present.
That is, NP‘O’S is re-interpreted as nP movement. The following pair shows the specific
vs non-specific order:
(38)
a.
ei du-To
bhai
this two-CLA
brother
‘these two brothers (here)’
b. ei bhaii du-To ti
‘these two brothers’
(non-specific; deictic)
(specific)
In the case of (38a) the base generated order of the Cla head duTo followed by the NP
bhai is manifested. However, notice that (38a) is purely deictic (see (32a) above). Deixis in
our theory is obtained by merging the Dem at [Spec,FP] and by moving a lower head to F
(see the discussion in connection with (32) and note 19). In (38a), therefore, I suggest that
duTo head moves to F as follows:
(39)
[FP ei [F duTo] [ClaP duTo [nP bhai]]]
In (38b) the NP is specific and therefore moves up to [Spec,ClaP] to check the strong
[specificity] feature of the Cla head. The Cla in this case does not head move and we
obtain non-locative deixis as desired (see (32b) in this connection). This movement of bhai
in this case does not obtain for us any difference that may exist between nP movement and
NP movement. To decide this, let us now consider the following where a full nP is
DP-internal NP movement
23
generated due to the presence of Poss.
(40)
a.
amar duTo
bhai
my two-CLA
brother
‘my two brothers’
b. amar bhai duTo
‘(these) two brothers of mine’
(contrastive)
(specific)
Similar to (38a), the order in (40a) is base-generated except for the movement of the Poss
to [Spec,DP]. Although this movement can be motivated by the scopal-MLC of Manzini
(1998), I will suggest that the movement of the Poss takes place via [Spec,ClaP] followed
by its movement to [Spec,DP] based on recent work by Ghomeshi (1997) which shows
that the Poss must take a presupposed/ definite object in the Indo-Iranian language
Persian. Object nouns in Persian may occur with the definite marker – , the indefinite
enclitic –i, or without any marker as shown in (41a). However, whenever a Poss is present,
the object NP must appear with the definite marker (41b).
(41)
a.
b.
ketab-o/ ketab-i/ ketab
xund-am
book-râ/ book-INDEF/ book read-1s
‘I read the book/ a book/ books’
ketab-e
jiân-o/ *jiân-i/ *jiân
xund-am
book-EZ
Jian-râ/ Jian-INDEF/ Jian read-1s
‘I read Jian’s book’
What we gain from this is the following. We can say that there is reason to believe that
Poss induces specificity effects and that perhaps they stop by an intermediate spec
position, possibly either/ both [Spec,ClaP] (for specificity) and [Spec,FP] position (for
Focus). This prediction is supported syntactically, since the Poss can use an intermediate
spec position as an escape hatch before moving out to [Spec,DP]. The final movement of
the Poss to its derived -- that is, the [Spec,DP] position -- is due to some feature [poss] in
D which attracts a Poss to its spec.
Coming back to (40a), notice that the phrase is contrastive even though there is no Dem
here. Therefore, although there is no Dem, the contrastive reading is obtained by head
movement of the Cla to F. This is shown in (42).
(42)
[DP amar [FP [F duTo] [ClaP amar [Cla duTo] [nP amar bhai ]]]]
24
Bhattacharya
If we compare the derivations in (39) and (42), we will notice that in (39) Cla→F results in
deixis and in (42) the same movement produces contrast. I will repeat the earlier
suggestion that in the former, deixis is obtained not only by moving the Cla to F but
additionally by merging the Dem at [Spec,FP]. In the case of (42), since there is no Dem,
Cla→F produces just a contrastive reading.21 Note that we are assuming that the F head
selected in (42) carries a strong feature of Focus.
In the case of (40b), however, this head movement does not take place as the Focus
feature is not strong. In effect, this means the NP gets a specific interpretation when the
Cla→F movement does not take place. The set of movements that derive (40b) consists of
a movement of the Poss to [Spec,DP] through [spec,ClaP] followed by the NP moving
leftwards to [Spec,ClaP] to induce specificity. However, in effect, this is similar to moving
the whole nP to [Spec,ClaP] first and then moving the Poss out to the higher Spec next. As
we have already established the need to have Poss and NP movement in the grammar,
there is no need to introduce another kind of NP movement. However, the number of steps
required for the derivation with nP movement is less than with Poss+NP movements.
Therefore, for economy reasons, I will consider movement due to specificity in non-KI
cases as nP movement:
(43)
[DP amar [FP [ClaP [SPEC amar bhai] [Cla duTo] tnP]]]
On the face of it, this seems like a strange result to fall out of the analysis. However, I will
suggest that the specificity effect is obtained by the “joint” action of something moving
into [Spec,ClaP] and the presence of some relevant head in Cla. This is identical to the
requirement that deixis is obtained by the joint action of merging the Dem in [Spec,FP]
and movement of a lower head into F, which, in turn was suggested to be part of a general
Focus Criterion of Brody (1990) (see note 19). This requirement of having both the Spec
and the Head filled to obtain certain syntactic effect is the property shared by deixis and
specificity.22
21
Notice that I am assuming that the specificity effect obtained by Poss movement in (40a) is
‘overshadowed’ by the contrastive requirement. I am not sure as to the semantic implications of such a
possibility.
22
This requirement runs counter to the suggestion in Giusti (1997) that only one of either the spec or the
head D needs to be occupied to obtain definiteness in DP. However, in many languages, including
Hungarian the requirement is just the opposite of what Giusti states. That is, both the [Spec,DP] and the D
must be filled to get a definite DP.
DP-internal NP movement
25
Let us now see if this redefinition of specificity works for the following examples
involving Dems.
(44)
a.
amar ei
du-To
bhai (non-specific; contrastive)
my this
two-CLA
brother
‘Lit: my these two brothers’
b. amar ei
bhaii du-To ti
(specific; non-locative deixis)
‘these two brothers of mine’
In (44a) there must be Cla→F movement since a contrastive reading is desired. However,
notice that Dem is also merged at the [Spec,FP] in (44a). Given our analysis of deixis as
the joint action of merging the dem at [Spec,FP] and moving Cla to F, (44a) must also
involve deixis. I will call this state of affairs ‘CONTRASTIVE DEIXIS’. This is shown in
(45a).
(45)
a.
[DP amar [FP ei [F duTo] [ClaP amar [Cla duTo] [nP amar bhai ]]]]
For the non-locative version in (44b), I suggest that the whole nP moves to [Spec,ClaP]
followed by the Poss moving to [Spec,DP]. However, since (44b) is specific, both the
spec and the head of ClaP must remain filled with the result that head movement of Cla
cannot take place. And since there is no Cla→F movement, the requirement that both the
spec and the head of FP must be filled is not met, with the result that the deixis obtained in
non-locative. This is shown below.
(45)
b.
[DP amar [FP ei [ClaP [SPEC amar bhai] [Cla duTo] tnP ]]]
3.3.3 NP movement. Now that we have charted the movement of the Poss all the way up
to the highest Spec in the DP, let us look at the data in connection with kinship terms.
Consider, e.g., the data in (35) repeated here:
(46)
a.
ei du-To
bhai
amar
this two-CLA
brother mine
‘these two brothers (of) mine, ....’
b. ei bhai
du-To amar
‘these two brothers (of) mine, ....’
....
(contrastive)
(non-locative)
26
Bhattacharya
Notice that the Poss in all cases so far carries the GEN Case marker. This would indicate
that in all cases, including the ones of KI where the Poss is stranded, D is generated since
that is the domain of GEN Case checking. We will therefore assume that the Poss satisfies
the [poss] feature of the D in KI cases covertly.
From (46) we can see that whenever the Dem is used in KI order, it is either contrastive
(as in 46a) or non-locative (as in 46b). Let see us now see how we derive (46). The
difference between the two versions, as we noted earlier, is that of specificity, the b
version, therefore, must involve movement to [Spec,ClaP] to induce the specificity effect
when the Cla head is filled:
(47)
[DP [FP ei [ClaP bhai [Cla duTo] [nP amar bhai ]]]]]
KI takes place in (46a) as well. KI is always obtained in our analysis by NP movement.
Movement to [Spec,ClaP] on the other hand will induce specificity if the Cla head also
remains filled by spell-out. Since the a. version is not specific, Cla head-moves to F (see
48). As a result of this movement we obtain contrastive deixis, as desired, for free.
(48)
[DP [FP ei [F duTo] [ClaP bhai [Cla duTo] [nP amar [NP bhai]]]]]
The deixis facts obtained in (46) fall out of the analysis that we have proposed based on a
difference between nP and NP movement.
4 Conclusions
The main findings of this paper are as follows. In section 1, I have shown that there are
clear cut cases of NP movement inside the DP in Bangla (Bengali) brought about by
specificity, in particular, a feature of [specificity] located at the Cla head which in this
language surfaces morphologically as the classifer.
In section 2, I have presented further evidence of DP-internal NP movement by looking
at new data on Kinship Terms. The proposed movement, Kinship Inversion, is also shown
to be triggered by [specificity] feature located at the Cla head. The base-generated
positions of the Possessive and the Demonstrative inside the Bangla DP are shown to be
the nP-shell and the specifier of a Focus-related head F respectively.
In section 3, the nature of deixis obtained inside the DP is investigated. I have shown
that NP movement results in ‘non-locative’ deixis. This in turn was explained on the basis
of a distinction between nP and NP movement. In particular, it was claimed that both
DP-internal NP movement
27
specificity and deixis are obtained by something moving into the head as well as the
specifier of a relevant projection (ClaP and FP respectively) – in effect, similar to the
Affect Criterion known in the literature to subsume the WH/NEG Criterion.
References
Bernstein, J. 1993. The Syntactic Role of Word Markers in Null Nominal Constructions. Probus 5.12: 5-38
 1997. Demonstratives and Reinforcers in Romance and Germanic Languages. Lingua 102: 87113
Bhattacharya, T. 1998a. Kinship Inversion in Bangla. Paper presented at 7th PG Conference in
Linguistics, University of Manchester, to be published in PLUM Series 7
Bhattacharya, T. 1998b. The nP-shell in Bangla. Paper presented at 19th Meeting of the South Asian
Linguistics Association, University of York.
Bhattacharya, T. (forthcoming). DP Internal Specificity in Bangla, in The Yearbook of South Asian
Languages and Linguistics 2, London: Sage Publication.
Brody, M. 1990. Some Remarks on the Focus Field in Hungarian. UCL Working Papers in
Linguistics 2: 201-225
Brugé, L. 1996. Demonstrative Movement in Spanish. A Comparative Approach. University of
Venice Working Papers in Linguistics 6.1: 1-53
Chomsky, N. 1995. The Minimalist Program. Cambridge, MA: MIT Press.
Collins, C. and Thráinsson, H. 1996. VP-internal structure and object shift in Icelandic. Linguistic
Inquiry 27.3: 391-444
Dasgupta, P. 1983. The Bangla Classifier /Ta/, its Penumbra, and Definiteness. Indian Linguistics 44:
11-26.
Diesing, M. 1992. Indefinites. Cambridge, MA: MIT Press.
Ghomeshi, J. 1997. Non-projecting Nouns in Persian. Natural Language and Linguistic Theory 15:
729-788.
Giorgi, A. and G. Longobardi. 1991. The Syntax of Noun Phrases. Cambridge: Cambridge
University Press.
Giusti, G. 1996. Is there a FocusP and TopicP in the Noun Phrase Structure? University of Venice
Working Papers 6.2: 105-128.
Giusti, G. 1997. The Categorial Status Of Determiners. The New Comparative Syntax, ed. L.
Haegeman: 94-113. Cambridge: Cambridge University Press.
Hamp, E. 1980. Unrecognised deixis in the Indo-European pronouns. Papers from the parasession
on pronouns and anaphors, CLS, eds. J. Kreiman and A. Ojeda: 147-150. Chicago, CLS.
Hazelkorn, L. T. 1983. The role of deixis in the development of Finno-Ugric grammatical morpheme.
Working Papers in Linguistics 27: 89-139.
Kayne, R.A. 1994. The Antisymmetry of Syntax. Cambridge, MA: MIT Press.
28
Bhattacharya
Kidwai, A. 1995. Binding and Free Word Order Phenomenon in Hindi and Urdu. Delhi: Jawaharlal
Nehru University diss.
Lehman, P. 1974. Proto-Indo-Eorpean Syntax. Austin and London: University of Texas Press
Longobardi, G. 1994. Reference and Proper Names: A Theory of N-movement in Syntax and Logical
Form. Linguistic Inquiry 25.4: 609-665.
Mahajan, A. 1990. The A/A-bar Distinction and Movement Theory.Cambridge, MA: MIT
dissertation.
Mallén, H. 1997. A Minimalist Approach to Concord in Noun Phrases. Theoretical Linguistics
23.1/2: 49-77
Manzini, R. 1998. A Minimalist Theory of Weak Islands. Syntax and Semantics: The Limits of
Syntax, vol 29, eds. Peter W. Culicover and Louise McNally 185-207. San Diego: Academic Press
Peterson, S. 1930. The inflection of Indo-European person nouns. Language 6: 164-193.
Rochemont, M.S. 1998. Phonological Focus and Structural Focus. Syntax and Semantics: The Limits
of Syntax, vol 29, eds. Peter W. Culicover and Louise McNally: 337-364. San Diego: Academic
Press
Rooth, M. 1985. Association with Focus. Amherst: University of Massachusetts diss.
Sheilds, K. 1994. The role of deictic particles in the IE personal pronoun system. Word 45.3: 307-315